Control Of Circular Cylinder Flow Research Articles

Dynamic feature-based deep reinforcement learning for flow control of circular cylinder with sparse surface pressure sensing

This study proposes a self-learning algorithm for closed-loop cylinder wake control targeting lower drag and lower lift fluctuations with the additional challenge of sparse sensor information, taking deep reinforcement learning (DRL) as the starting point. The DRL performance is significantly improved by lifting the sensor signals to dynamic features (DFs), which predict future flow states. The resulting DF-based DRL (DF-DRL) automatically learns a feedback control in the plant without a dynamic model. Results show that the drag coefficient of the DF-DRL model is 25 % less than the vanilla model based on direct sensor feedback. More importantly, using only one surface pressure sensor, DF-DRL can reduce the drag coefficient to a state-of-the-art performance of approximately 8 % at Reynolds number $(Re) = 100$ and significantly mitigates lift coefficient fluctuations. Hence, DF-DRL allows the deployment of sparse sensing of the flow without degrading the control performance. This method also exhibits strong robustness in flow control under more complex flow scenarios, reducing the drag coefficient by 32.2 % and 46.55 % at $Re =500$ and 1000, respectively. Additionally, the drag coefficient decreases by 28.6 % in a three-dimensional turbulent flow at $Re =10\,000$ . Since surface pressure information is more straightforward to measure in realistic scenarios than flow velocity information, this study provides a valuable reference for experimentally designing the active flow control of a circular cylinder based on wall pressure signals, which is an essential step toward further developing intelligent control in a realistic multi-input multi-output system.

Surrogate model-based deep reinforcement learning for experimental study of active flow control of circular cylinder

The flow around a circular cylinder is a classical problem in fluid mechanics, and the reduction of drag and lift has been a long-standing research focus in flow control. In this study, we apply deep reinforcement learning (DRL) to intelligently determine suction flow rate on a circular cylinder model in wind tunnel, aiming to minimize aerodynamic forces while considering energy dissipation efficiency. However, DRL has been criticized for its low data utilization rate and long training period, leading to high experimental training cost. To address these issues, this study employs a surrogate model to optimize the reward function and hyperparameters, and this method is called SM-DRL. This SM-DRL method efficiently expedites the DRL training process, significantly reducing the experimental training cost. In addition, DRL training was conducted in a variable flow field, and the robustness of the obtained DRL model was tested. The results indicate that the DRL agent can determine the optimal control strategy, i.e., automatically select the optimal suction flow rate in terms of the incoming wind velocity, resulting in a significant reduction in lift fluctuations. For Reynolds number of 1.65×104, the reduction in lift fluctuations of the circular cylinder exceeds 50%.

Experimental study on passive flow control of circular cylinder via perforated splitter plate

Present experimental investigation aims to reduce the shedding of vortex in the near wake region of a circular cylinder using a perforated splitter plate. Perforated plates were placed in the wake region of the cylinder and aligned with the streamwise direction. The length of the plates was equal to the diameter of the cylinder. Different plate porosities and locations were examined and obtained results were compared to the baseline cylinder. Flow measurements downstream of the cylinder were performed in a water channel by employing a particle image velocimetry technique (PIV) at a Reynolds number of Re=5x103. It is observed that the effect of the porosity on the flow characteristics of the cylinder depends on the location of the plate. The strength of shear layers and flow fluctuations in the near wake region of the cylinder are considerably diminished by the perforated splitter plate. It is found that the porosity of e=0.3 is the most effective control element for gap ratio of G/D=0.5. On the other hand, proper gap ratio is determined as G/D=2 for porosity of e=0.7. It is concluded in the present study that the perforated splitter plate could be used as alternative passive flow control technique in order to reduce vortex shedding of the cylinder.

Flow control of circular cylinder with ice-induced structure

Flow control of circular cylinder with ice-induced structure

Control of circular cylinder flow via bilateral splitter plates

We carry out wind tunnel investigations to study the flow of a circular cylinder modified with two rigid splitter plates hinged along its stagnation points. The equal-sized and symmetrically placed splitter plates are both parallel to the incoming airflow, and their single-sided length in the streamwise direction varies from 0 to 2.0D (where D is the cylinder diameter). The wind tunnel experiments are conducted at the Re of 3.33 × 104. In addition to bilaterally arranged plates, two other configurations of splitter plates, i.e., front-plate-only and rear-plate-only, are also investigated. By employing the sectional measurement of surface pressure in the midspan slice, we evaluate typical aerodynamic parameters, including pressure distribution, instantaneous drag and lift forces, frequency spectra of the unsteady lift forces, mean drag, and root-mean-square lift coefficients acting on the cylindrical test models. A particle image velocimetry (PIV) system is used to visualize and quantify the vortex shedding process and the dynamic interactions of the natural and modified cylinders. Experimental results of the surface pressure measurement and PIV measurement results are then combined to reveal the effects of rigid plates with different configurations (bilateral, front-only, and rear-only) on the circular cylinder flow.

Active control of circular cylinder flow with windward suction and leeward blowing

This study is an experimental investigation of the effectiveness and mechanism of a concept for bluff-body control characterized by combined windward suction and leeward blowing (WSLB). Wind tunnel tests are performed at a subcritical Reynolds number (Re) of 3.33 × 104. Open-loop WSLB is used to control the flow around a cylindrical test model. Distributed suction and blowing nozzles are symmetrically arranged at the front and rear stagnation points. Steady suction and blowing are implemented simultaneously on both surfaces of the cylinder. WSLB control is characterized by the dimensionless momentum of the suction/blowing relative to the incoming airflow. Instantaneous pressure distributions on the midspan of the cylinder surface for the baseline and controlled cases are measured to quantify the modifications of WSLB control to the aerodynamic coefficients of the cylinder. In addition to surface pressure measurements, a particle image velocimetry (PIV) system is used to describe the wake flow structures of the baseline and controlled cylinders. Experimental results demonstrate that WSLB control decreases sectional drag at the midspan and reduces the fluctuating amplitudes of dynamic wind loads acting on the cylinder. The Strouhal number to characterize the vortex shedding frequencies of the controlled cylinders is also found to deviate from that of the natural cylinder. PIV measurement results show that the active blowing positioned at the leeward stagnation point forms a pair of vortices into the cylinder wake that modify the original vortex shedding process. As the blowing vortices convect into the wake, they stretch the unsteady shear flows from the upper and lower sides of the cylinder, increase the vortex-formation length and push the alternative vortex shedding further downstream. It is also shown that a steady and symmetric perturbation imposed on the periodic cylinder flow can reduce the von Karman vortices and modify the mode of wake vortex shedding significantly. With active control of windward suction and leeward blowing (WSLB), blowing positioned at the leeward stagnation point forms a pair of vortices in the cylindrical wake that modifies the typical vortex shedding process. As the blowing vortices convect downstream, they contribute to detaching the unsteady shear layers from the cylinder. It is shown that a steady and symmetric perturbation imposed on the periodic cylinder flow can lead to a symmetric mode of wake vortex shedding.

Control of circular cylinder flow using distributed passive jets

A wind tunnel study has been carried out to investigate flow control around a hollow circular cylinder using passive jets driven by naturally occurring pressure differences. Flow enters the cylinder through spanwise holes along the stagnation line and exits through a spanwise distribution of holes at$\pm 65^{\circ }$. The diameter of the entry and exit holes were 1 % and 0.5 % of the cylinder diameter, respectively. Reynolds numbers were at the upper end of the subcritical regime and ranged from$3\times 10^{4}$to$2.8\times 10^{5}$. Jet spacings of 10 % and 20 % of the cylinder diameter were investigated, and the ratio of the average jet exit velocity to the cross-flow velocity at the boundary layer edge was found to rise to approximately 0.35 and 0.4, respectively, above a Reynolds number of$1.5\times 10^{5}$. Findings based on using the surface oil flow technique revealed a repeating, organised cellular pattern downstream of adjacent jet exit holes consisting of a primary counter-rotating vortex pair structure, followed by a secondary weaker pair. Downstream of adjacent exit holes, and centred midway between them, there exists a separation bubble which delays final flow separation compared with the flow directly downstream of a jet. The variation in the angular position of boundary layer separation across the span had the effect of suppressing von Kármán vortex shedding. This resulted in a drag coefficient, at the upper end of the Reynolds-number range studied, 14.5 % lower than that found using trip wires to initiate boundary layer transition.

Active control of circular cylinder flow by affiliated rotating cylinders

This study puts forward an active control method for circular cylinder flow by placing two small affiliated rotating cylinders adjacent to the main cylinder, and their effects on the drag and lift forces acting on the main cylinder as well as the heat transfer effectiveness are numerically investigated. According to the diameter of the main cylinder the Reynolds number is chosen as Re=200. The well-proven finite volume method is employed for the calculation. The code is validated by comparing the present computed results of flow passing an isolated rotating cylinder with those available from the literature. To describe the present control model, two parameters are defined: the rotation direction of the two small cylinders (including co-current rotation and counter-current rotation) and the dimensionless rotation rate α. In the simulation, the rotation rate α varies from 0 to 2.4. The results indicate that the optimum rotation direction of the subsidiary cylinders, which is beneficial to both drag reduction and heat transfer enhancement, is the co-current rotating (the upper affiliated cylinder spins clockwise and the lower affiliated cylinder spins counter-clockwise). We observe noticeable suppression of the vortex shedding and favorable reduction of the fluid forces acting on the main cylinder as the rotation rate increases. Besides, the pressure and viscous components of the drag force are analyzed. Energy balance between energy cost for activating the rotating cylinders and energy saving by the momentum injection is considered. In addition, the influence of the affiliated rotating cylinders on heat transfer is also investigated. The average Nusselt number is found to increase with the rotation rate.

Flow control of circular cylinder with a V-grooved micro-riblet film

Flow structure around a circular cylinder with a V-grooved micro-riblet is investigated experimentally. The results are compared with that of a smooth cylinder having the same diameter. A flexible V-shaped micro-riblet with peak-to-peak spacing of 300 μ m is made using a MEMS fabrication process of PDMS (Polydimethylsiloxane) replica. The flexible micro-riblet is attached on a circular cylinder with which grooves are aligned with the streamwise flow direction. Drag force acting on the cylinder is measured for Reynolds numbers based on a cylinder diameter ( D = 18 mm ) in the range Re D = 2.5 × 10 3 – 3.8 × 10 4 . At Re D = 3.6 × 10 3 ( U 0 = 3 m / s ) , the V-grooved micro-riblet cylinder reduces drag coefficient by 7.6%, compared with a cylinder with smooth PDMS surface. However, it increases drag coefficient about 4.2% at Re D = 3.6 × 10 4 ( U 0 = 30 m / s ) . Flow field around the micro-riblet cylinder is measured by using a 2-frame PIV velocity field measurement technique. Several hundreds instantaneous velocity fields are ensemble-averaged to get the spatial distributions of turbulent statistics including turbulence intensities and turbulent kinetic energy. For the case of drag reduction at Re D = 4.8 × 10 3 , the vortex formation region behind the V-grooved MRF cylinder is reduced about 10%, compared with the smooth cylinder due to enhanced entrainment of ambient inviscid fluid into the wake region. In addition, the total number of secondary vortices located inside the near wake region is decreased about 20%.

Control of circular cylinder flow by the use of dimples

Measurements are reported of the drag coefficient and Strouhal number of a dimpled circular cylinder over the Reynolds number range from 2 x 104 to 3 x 10s. The ratio of the depth of the dimples to the diameter of the cylinder is 9xlO~ 3. In common with sand-roughened cylinders, the dimpled cylinder has a lower critical Reynolds number than a smooth cylinder. After the drag coefficient minimum, the CD does not rise to the high values that are typical of cylinders with sand roughness but is found to be closer to that for a smooth cylinder. Over a Reynolds number range from about 4xl04 to 3xl05, a dimpled circular cylinder has a lower drag coefficient than a smooth cylinder.